Predicting the optimal noise strength for solving optimization problems with analog Ising machines

TL;DR

This paper explores how injecting optimized noise levels can significantly enhance the performance of analog Ising machines in solving MaxCut problems, reducing the need for extensive parameter tuning.

Contribution

It introduces a method to predict optimal noise strength based on problem parameters, improving success rates and time-to-solution without extensive tuning.

Findings

Optimized noise injection improves TTS by several orders of magnitude.

Predicting noise strength based on problem connectivity eliminates costly tuning.

The approach makes analog Ising machines competitive with state-of-the-art methods.

Abstract

Analog Ising machines are dedicated hardware solvers designed to solve NP hard optimization problems. However, the global optimum is often not found as the system gets stuck in local minima. While several strategies exist to increase the chance of escaping local minima, often these methods needs extensive parameter tuning. In this work, we investigate the injection of large noise as a scheme on its own and in combination with annealing to improve the success rate and the time-to-solution (TTS) of analog Ising machines for MaxCut problems. We demonstrate that optimizing the noise improves the TTS by several orders and makes both approaches competitive with the state-of-the-art, such as chaotic amplitude control. Moreover, we are able to predict a good noise value based on the problem connectivity and coupling strength, eliminating the need for costly parameter optimization.